Artificial Intelligence and Lights-Out Court Document Processing
Tesla makes self-driving cars, Amazon can predict what you are interested in purchasing, Starbucks now even knows what your coffee order will be when you walk into their store – all possible due to artificial intelligence (AI) technology in use today. So why are we spending millions of hours each year reviewing scanned and e-filed court documents and updating data in case management systems?
AI software can automate court document workflows, review and validate filings, and perform automatic data entry. AI software bots can perform as virtual employees, providing 24x7x365 “lights-out” document processing. Any case management system, any document and any workflow, all without writing a single line of code. AI software and bots can even automate the exchange of data between disparate systems where the only current methodology is tedious, error-prone, and time-consuming manual data entry.
What then, is AI software? AI software is divided into one of two categories, General AI and Narrow AI. General AI focuses on simulated human intelligence, resulting in fabulous machines which have all our senses and more. Some examples are self-driving cars, IBM Watson, and other deep learning technologies. Narrow AI, on the other hand, encompasses technologies which are able to perform specific tasks as well as, or better than, we humans can. Examples of such are electronic stock trading, facial recognition, weather forecasting, and of importance to courts: unstructured document analysis.
Narrow AI solutions are often constructed using state-of-the-art machine learning algorithms. In 1959, Arthur Samuel, a pioneer in the field of AI, defined machine learning as a field of study that gives computers the ability to learn without being explicitly programmed. Given examples of data, machines teach themselves and make predictions within specific knowledge domains. Machine learning consists of two distinct phases. The initial phase, supervised learning, provides data for algorithms to train on, producing knowledge that is then reused to perform specific tasks. The secondary, or reinforced learning phase, uses production processing results to automatically refine knowledge, allowing the system to grow smarter over time.
Narrow AI addresses four areas of automating court document workflows – automatic document classification, document separation, data extraction and advanced redaction. Traditional legacy-based solutions made use of arcane regular expressions. They are complex, static, expensive to develop, and fragile, as when they encounter new document types or data fields, produce poor results. State-of-the-art machine learning makes use of text decomposition, natural language processing, and both spatial and linguistic analysis. AI solutions are flexible, extensible, self-improving, and highly accurate. Practical applications of AI for court document processing are as follows.
- Error-free e-filed documents. Attorneys file drop their documents into an e-filing portal, then verify the docket classifications, extracted data, and redactions all based upon actual document contents.
- Unattended document intake for case management systems. Software analysis eliminates Clerk manual acceptance and review of documents and data, while robots perform data entry.
- Refinement of docket coding. Generalized e-portal codes enhanced to consistent specific local jurisdiction coding.
- Automated secondary data entry. Eliminate docket-specific secondary reviews and data entry workflows.
- Ad-Hoc information extraction and data mining. Perform analytics directly from case document contents and not just what is stored in a case management system.
- Automation of non-interfaced systems
AI is upon us and is a beneficial disruptive technology. AI assists attorneys in isolating relevant records during e-discovery; AI predicts outcomes of cases and assists attorneys in developing case strategies; AI powers web based chat bot conversations. AI will continue to grow and provide a strategic advantage to those that embrace its superhuman capabilities.
This post is also featured in the November 2017 Issue of Courts Today Magazine, pg.8.